grakel.LovaszTheta

class grakel.LovaszTheta(n_jobs=None, normalize=False, verbose=False, random_state=None, n_samples=50, subsets_size_range=(2, 8), max_dim=None, base_kernel=None)

Lovasz theta kernel as proposed in [JJDB14].

Parameters

X,Yvalid-graph-format

The pair of graphs on which the kernel is applied.

n_samplesint, default=50

The number of samples.

subsets_size_rangetuple, len=2, default=(2,8)

(min, max) size of the vertex set of sampled subgraphs.

random_stateRandomState or int, default=None

A random number generator instance or an int to initialize a RandomState as a seed.

base_kernelfunction (np.1darray, np.1darray -> number), default=None

The applied metric between the lovasz_theta numbers of subgraphs. If None \(f(x,y) = x*y\)

max_dimint, default=None

The maximum graph size that can appear both in fit or transform. When None, max_dim is calculated based on the size of the biggest graph on fit. This can lead to a crash in case a graph appears in transform with size bigger than in fit.

Attributes

d_int,

The maximum matrix dimension of fit plus 1. Signifies the number of features assigned for lovasz labelling.

random_state_RandomState

A RandomState object handling all randomness of the class.

Methods

diagonal()	Calculate the kernel matrix diagonal of the fit/transformed data.
fit(X[, y])	Fit a dataset, for a transformer.
fit_transform(X)	Fit and transform, on the same dataset.
get_params([deep])	Get parameters for this estimator.
initialize()	Initialize all transformer arguments, needing initialization.
pairwise_operation(x, y)	Lovasz theta kernel as proposed in [JJDB14].
parse_input(X)	Parse and create features for lovasz_theta kernel.
set_params(**params)	Call the parent method.
transform(X)	Calculate the kernel matrix, between given and fitted dataset.

Initialise a lovasz_theta kernel.

Attributes

X

Methods

diagonal()	Calculate the kernel matrix diagonal of the fit/transformed data.
fit(X[, y])	Fit a dataset, for a transformer.
fit_transform(X)	Fit and transform, on the same dataset.
get_params([deep])	Get parameters for this estimator.
initialize()	Initialize all transformer arguments, needing initialization.
pairwise_operation(x, y)	Lovasz theta kernel as proposed in [JJDB14].
parse_input(X)	Parse and create features for lovasz_theta kernel.
set_params(**params)	Call the parent method.
transform(X)	Calculate the kernel matrix, between given and fitted dataset.

__init__(n_jobs=None, normalize=False, verbose=False, random_state=None, n_samples=50, subsets_size_range=(2, 8), max_dim=None, base_kernel=None)

Initialise a lovasz_theta kernel.

Bibliography

JJDB14(1,2,3)

Fredrik Johansson, Vinay Jethava, Devdatt Dubhashi, and Chiranjib Bhattacharyya. Global graph kernels using geometric embeddings. In Proceedings of the 31st International Conference on Machine Learning, 694–702. 2014.